Qwen 3.7 Max vs DeepSeek V4: Benchmark Comparison
Qwen 3.7 Max has the higher score on 8 of 15 shared benchmarks; DeepSeek V4 leads on 7.
The largest observed score gap is 13.13 pts on Code Migration , where DeepSeek V4 leads.
Reported ±1 standard-error ranges overlap on 7 of 15 shared benchmarks with comparable uncertainty data. This is not a pairwise statistical significance test.
Shared benchmark results
Scores and reported standard errors come from the latest visible Vals benchmark version available for both models.
| Benchmark | Qwen 3.7 Max | DeepSeek V4 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 1.67% ±0.83 | 3.75% ±1.43 | 2.08 pts | Reported ±1 SE ranges overlap |
| Legal Research Bench | 25.48% ±3.03 | 23.08% ±2.93 | 2.40 pts | Reported ±1 SE ranges overlap |
| LegalBench | 84.91% ±0.41 | 80.32% ±0.47 | 4.59 pts | Reported ±1 SE ranges do not overlap |
| EMB | 57.04% ±2.85 | 51.62% ±3.05 | 5.42 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 47.78% ±0.73 | 44.08% ±0.65 | 3.69 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 47.41% ±3.24 | 58.50% ±3.13 | 11.09 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 75.31% ±0.84 | 72.08% ±0.88 | 3.23 pts | Reported ±1 SE ranges do not overlap |
| MedCode | 38.75% ±2.20 | 40.45% ±2.12 | 1.70 pts | Reported ±1 SE ranges overlap |
| MedScribe | 79.40% ±1.91 | 75.14% ±2.00 | 4.25 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 90.15% ±1.50 | 89.39% ±1.65 | 0.76 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 89.31% ±0.30 | 87.25% ±0.34 | 2.06 pts | Reported ±1 SE ranges do not overlap |
| Code Migration | 13.07% ±2.86 | 26.20% ±4.04 | 13.13 pts | Reported ±1 SE ranges do not overlap |
| LiveCodeBench | 87.06% ±0.97 | 87.48% ±0.95 | 0.43 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 68.80% ±2.07 | 77.40% ±1.87 | 8.60 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 47.67% ±4.63 | 49.93% ±4.77 | 2.26 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Qwen 3.7 Max average | DeepSeek V4 average |
|---|---|---|
| Legal | 37.35% | 35.72% |
| Finance | 56.88% | 56.57% |
| Healthcare | 59.07% | 57.80% |
| Academic | 89.73% | 88.32% |
| Coding | 54.15% | 60.25% |
Cost and latency
Cost per test appears only where the benchmark reports it for both models. Latency is the measured completion time for that benchmark.
| Benchmark | Qwen 3.7 Max cost | DeepSeek V4 cost | Qwen 3.7 Max latency | DeepSeek V4 latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $1.58 | $0.74 | 32m03s | 6m42s |
| Legal Research Bench | $1.40 | $0.68 | 13m29s | 17m05s |
| LegalBench | N/A | N/A | 11.85s | 3m04s |
| EMB | $2.86 | $0.80 | 32m37s | 13m42s |
| Finance Agent (v2) | $0.55 | $0.88 | 10m35s | 25m26s |
| Tax Agent Bench | $0.44 | $0.72 | 3m30s | 20m45s |
| TaxEval v2 | N/A | N/A | 57.68s | 3m10s |
| MedCode | N/A | N/A | 28.75s | 6m23s |
| MedScribe | N/A | N/A | 108.05s | 5m46s |
| GPQA Diamond | N/A | N/A | 2m07s | 15m00s |
| MMLU Pro | N/A | N/A | 29.15s | 9m38s |
| Code Migration | $1.93 | $1.07 | 25m32s | 28m58s |
| LiveCodeBench | N/A | N/A | 3m12s | 11m56s |
| SWE-bench | $3.23 | $0.44 | 16m16s | 10m35s |
| Vibe Code Bench v1.1 | $11.40 | $2.21 | 1h18m | 56m59s |
Results available only for Qwen 3.7 Max
None.
Results available only for DeepSeek V4
- Vals Index
- ProofBench v1.1
- ProgramBench
- SkillsBench
- Terminal-Bench 4.0
- Public Benefits Bench v1.1